The Eutherian Armcx Genes Regulate Mitochondrial Trafficking in Neurons and Interact with Miro and Trak2

ARTICLE

Received 13 Sep 2011 | Accepted 10 Apr 2012 | Published 8 May 2012 DOI: 10.1038/ncomms1829 The Eutherian Armcx genes regulate mitochondrial trafficking in neurons and interact with Miro and Trak2

Guillermo López-Doménech1,*, Román Serrat1,*, Serena Mirra1, Salvatore D’Aniello2,†, Ildiko Somorjai2,†, Alba Abad3, Nathalia Vitureira1, Elena García-Arumí4, María Teresa Alonso5, Macarena Rodriguez-Prados5, Ferran Burgaya1, Antoni L. Andreu4, Javier García-Sancho5, Ramón Trullas3, Jordi Garcia-Fernàndez2 & Eduardo Soriano1

Brain function requires neuronal activity-dependent energy consumption. Neuronal energy supply is controlled by molecular mechanisms that regulate mitochondrial dynamics, including Kinesin motors and Mitofusins, Miro1-2 and Trak2 proteins. Here we show a new protein family that localizes to the mitochondria and controls mitochondrial dynamics. This family of proteins is encoded by an array of armadillo (Arm) repeat-containing genes located on the X chromosome. The Armcx cluster is unique to Eutherian mammals and evolved from a single ancestor gene (Armc10). We show that these genes are highly expressed in the developing and adult nervous system. Furthermore, we demonstrate that Armcx3 expression levels regulate mitochondrial dynamics and trafficking in neurons, and that Alex3 interacts with theM Kinesin/ iro/Trak2 complex in a Ca2 + -dependent manner. Our data provide evidence of a new Eutherian-specific family of mitochondrial proteins that controls mitochondrial dynamics and indicate that this key process is differentially regulated in the brain of higher vertebrates.

1 Developmental Neurobiology and Regeneration Lab, IRB Barcelona;Parc Cientific de Barcelona, Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED, ISCIII), Department of Cell Biology, University of Barcelona, Barcelona E-08028, Spain. 2 Department of Genetics, Faculty of Biology, and Institute of Biomedicine (IBUB), University of Barcelona, Barcelona E-08028, Spain. 3 Department of Neurobiology, Institut d′Investigacions Biomèdiques de Barcelona, CSIC, IDIBAPS, Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED, ISCIII), Barcelona E-08036, Spain. 4 Department de Patologia Neuromuscular i Mitocondrial, Hospital Universitari Vall d’Hebron Institut de Recerca (VHIR), Centro de Investigación Biomédica en Red de Enfermedades Raras (CIBERER), ISCIII, Barcelona 08035, Spain. 5 Department of Molecular and Cellular Physiology, Instituto de Biología y Genética Molecular (IBGM), Universidad de Valladolid & CISC, 47003 Valladolid, Spain. *These authors contributed equally to this study. †Present address: Department of Cellular and Developmental Biology, Stazione Zoologica Anton Dohrn, Villa Comunale 80121, Napoli, Italy (S.D); Centre for Organismal Studies, Im Neuenheimer Feld 230, University of Heidelberg, 69120 Heidelberg, Germany (I.S.). Correspondence and requests for materials should be addressed to E.S. (email: [email protected]). nature communications | 3:814 | DOI: 10.1038/ncomms1829 | www.nature.com/naturecommunications © 2012 Macmillan Publishers Limited. All rights reserved. ARTICLE nature communications | DOI: 10.1038/ncomms1829

itochondrial function is vital for neuronal viability, neu- signal peptide with a potential transmembrane domain), nuclear rotransmission and neural integration1–6. Synaptic neu- localization signals and putative mitochondrial targeting sequences Mrotransmission and integration events in nervous tissue (Fig. 1b), in agreement with our data indicating that Armcx genes take place at long distances from neuronal cell bodies (dendrites are targeted to. Taken together, these observations suggest that these and axons). Consequently, to ensure energy production in distal proteins have related physiological or cellular functions. Moreover, compartments, neurons orchestrate exquisite mechanisms that unusually, all Armcx-coding sequences are contained in a single regulate the transport and localization of mitochondria at active exon, whereas the Armc10 is a typical, multi-exon-containing gene synaptic sites7–11. Mitochondrial trafficking in neurons is mediated with the coding sequence split into at least eight exons. by Kinesin and Dynein motors and by several GTPases and adaptor On the basis of the above data, it is likely that the entire Armcx proteins, including Miro1/2, Trak2 or Mitofusins12–16. Mutations cluster originated in early Eutherian evolution by retrotransposition in genes encoding mitochondrial trafficking proteins often lead to of an Armc10 messenger RNA, and then by consecutive tandem neurological and neurodegenerative diseases17–19. duplication events in a rapidly evolving region of the Xq chromo- Members of the Alex protein family (Alex1–3; for arm-contain- some23, which also includes the Armcx-related G-protein-coupled ing protein lost in epithelial cancers linked to the X chromosome) receptor Gasp1–3 genes24. The precise time point for the genesis were initially described as putative tumor-suppressor genes, as their of the Armcx cluster, when key innovations of the mammalian expression is reduced in several epithelial-derived carcinomas, group took place, namely the origin of a well-developed placenta, including lung, prostate, colon and pancreas cancers20. While Alex1 or the increase in complexity of the central nervous system and and 2 are widely expressed, Alex3 is found mainly in the nervous the enlargement of the neocortex25, suggest that this cluster has system. Previously, we characterized Alex3 as a gene preferentially functions related to the above processes. Hence, the origin of the expressed in the upper layers of the developing cerebral cortex21. Armcx cluster may well have been linked to the enlargement of the A recent report described Alex3 as a Sox10-interacting protein neocortex, as marsupials are devoid of the Armcx cluster but have a that localizes in the mitochondria of OBL21 cells, and suggested a primitive placenta. novel signalling cascade between the mitochondria and the nucleus We next analysed the expression of Armcx genes during embry- through a Sox10/Alex3 protein complex22. Here, we describe the onic development and in the adult brain. At embryonic stages, evolutionary history of the Armcx cluster and provide evidence that Armc10 and Armcx3–6 genes were highly expressed in the devel- these genes encode a new family of proteins that regulates mito- oping neural tissues, neural crest derivatives and hind limbs, in chondrial dynamics and that is specific to Eutherian mammals. addition to other tissues that were specific for each gene (Supple- Furthermore, we show that Alex3 protein levels affect the dynam- mentary Fig. S5). We also found that Armc10 and Armcx1–6 genes ics and trafficking of mitochondria in neurons through interac- were widely expressed in the adult nervous tissue, especially in the tion with the Kinesin/Miro/Trak2 complex in a Ca2 + -dependent forebrain, including the cerebral cortex, hippocampus and thala- manner. We conclude that the Armcx/Arm10 genes encode a mus (Fig. 1c). We thus conclude that the Armcx genes encode for a new family of proteins that controls mitochondrial dynamics and Eutherian-specific set of proteins that is highly expressed in transport in neurons. neuronal tissue.

Results Armcx and Armc10 genes encode for mitochondrial proteins. The Armcx cluster originated during the evolution of placentals. To study the subcellular localization of Armcx gene products, To gain insight into the function of Armcx genes, we surveyed the HEK293AD cells were transfected with complementary DNAs mouse and human genomes for similar sequences. Interestingly, a full (cDNAs) encoding full-length Armcx proteins fused to EGFP set of related genes map to the same region of the human (Xq22.1) or myc tags. In agreement with the protein domain predictions and mouse (X56/57cM) X chromosomes, with the sole exception of reported above, all the Alex proteins tested (Alex1, 2, 3 and 6) were the Armc10/SVH gene, which maps to the seventh human and the targeted mainly to mitochondria, which were visualized by cellu- fifth mouse chromosome23,24. This unusual organization prompted lar staining with MitoTracker (Fig. 1c; and Supplementary Fig. S6). us to analyse further the evolutionary genesis of this gene family Nuclear staining was also occasionally seen in individual cells. (Fig. 1a). Six homologues, named Armcx1–6 and a pseudogene (Fig. Transfection with a cDNA encoding Armc10 yielded an identical 1a), are arrayed in a single cluster in a portion of the X chromosome, pattern of mitochondrial staining (Fig. 1c). Many cells transfected which was proposed to be specific to placental mammals23,24. We with Armcx1, 2, 3 and 6 and Armc10 cDNAs showed an altered then surveyed all available vertebrate genomes and noted the absence mitochondrial network, in which mitochondria were aggregated of Armcx genes in non-Eutherian mammals, namely monotremes in the perinuclear region (Fig. 2a and Supplementary Fig. S6). We and marsupials, and in non-mammalian vertebrates, and its conclude that the Armc10/Armcx cluster encodes for a new fam- consistent presence in all Eutherian mammals, including the earliest ily of proteins targeted to mitochondria and possibly regulates the branching armadillo (Supplementary Figs S1–S3). Hence, the Armcx dynamics and distribution of these organelles. cluster originated at the beginning or early radiation of the most To gain insight into the functional role of Alex proteins, we advanced Eutherian mammals (Fig. 1a). Furthermore, the closest focussed on Alex321. We generated antibodies against a synthetic phylogenetic relative to Armcx genes, the Armc10 gene, is present peptide corresponding to the Alex3 sequence. Following transfec- in a single copy in all the vertebrate genomes analysed, including tion of Alex3 in HEK293AD cells, these antibodies recognized this the descendants of the earliest branching mammals (monotremes protein in western blots (WBs) and immunofluorescence Fig.( 2a). and marsupials) and Eutheria. In addition, phylogenetic analyses A similar band was detected in WBs of brain lysates (Supplementary positioned the Armc10 gene as the sister group to all Armcx genes Fig. S7). The regional and cellular pattern of Alex3 immunostaining (Supplementary Fig. S1 and Supplementary Table S1), and in silico in brain sections was identical to that found in in situ hybridization prediction emphasized the presence of up to six Arm-like repeats analyses (Fig. 2b,c). Thus, Alex3 protein was widely expressed in in the carboxy (C)-terminal region of all proteins constituting the most brain regions and neurons; Alex3 protein expression peaked uncharacterized DUF634 domain (www.ebi.ac.uk/interpro/Display at postnatal stages (P0–P21) and decreased slightly thereafter. In IproEntry?ac=IPR006911) (Fig. 1b). All the predicted Arm domains transfected HEK293AD cells, Alex3 immunostaining colocalized include the characteristic three α-helices rich in hydrophobic with mitochondria, where Alex3 signals appeared to wrap indi- amino acids (Supplementary Fig. S4). Furthermore, most Armcx vidual mitochondria (Fig. 2d; Supplementary Fig. S8a,b). Immuno- proteins include signal peptides (such as an amino (N)-terminal gold electron microscopy confirmed that Alex3 protein surrounded

nature communications | 3:814 | DOI: 10.1038/ncomms1829 | www.nature.com/naturecommunications © 2012 Macmillan Publishers Limited. All rights reserved. nature communications | DOI: 10.1038/ncomms1829 ARTICLE

a Eutherians Armcx cluster

Mouse X 56/57 cM, Human Xq22.1

4 1 7 6 3 2 5 Eutherians Armc10 Armcx4 Armcx1 Armcx6ps Armcx6 Armcx3 Armcx2 Armcx5 Mouse 5, 9, 97cM Alex1 Armcx6-like Alex3 Alex2 Human 7q22.1

Marsupials Armc10 Monotremes Armc10 Mammals Birds/reptiles Armc10 Amphibians Armc10 Teleost fishes Armc10 Bony vertebrates

b 306 Armc10

456 Armcx1 Signal peptide 784 Armcx2 Mitochondrial targeting 379 site Armcx3 Nuclear localization 2356 signal Armcx4 Arm-like repeats 606 Armcx5 DUF463 domain 301 Armcx6 189 Armcx6ps

c Armc10 Armcx1 Armcx2 Armcx6

h P21

t Hoechst / racker MitoT / myc

Figure 1 | Evolution and expression of the Armcx gene cluster. (a) Evolutionary origin of the Armcx cluster in Eutherians, through retrotransposition of the autosomal Armc10 gene (curved arrow) and tandem duplication in the ancestral Eutherian X chromosome. (b) Schematic representation of the Armc10/Armcx family. They all have a highly conserved C-terminal domain (DUF463) consisting of up to six Armcx or Arm10 genes-like tandem repeats. In contrast, the N-terminal region is unrelated among members and highly variable in length and sequence, although various members share several common features, such as an N-terminal signal peptide with a potential transmembrane domain, a putative mitochondrial targeting sequence and a nuclear localization signal. (c) Expression of Armcx messenger RNAs in mouse brain and subcellular localization of Armcx proteins. The upper panels show in situ hybridization demonstrating high expression of Armc10, Armcx1, Armcx2 and Armcx6 genes in the hippocampus, neocortex and thalamus. The lower panels show mitochondrial targeting of Armc10, Armcx1, Armcx2 and Armcx6 proteins after transfection of Myc-tagged cDNAs in HEK293AD cells. Mitochondria were stained with the mitochondrial marker MitoTracker (red) and nuclei were stained with Hoechst (blue). Scale bars: upper panels 1 mm; and lower panels 10 µm.

nature communications | 3:814 | DOI: 10.1038/ncomms1829 | www.nature.com/naturecommunications © 2012 Macmillan Publishers Limited. All rights reserved. ARTICLE nature communications | DOI: 10.1038/ncomms1829

a pcDNA Alex3 Alex3/Hoechst g Alex3 Hoechst Merge

42 WB: 39 α-Alex3 C8

b I c h Alex3 MitoTracker Merge II-IV I V II-IV VI V CA1 VI CA1 DG CA3

DG CA3 Ad Ad d Alex3 MitoTracker Merge i Alex3 Actin Merge

e pcDNA f Alex3 j Alex3 MitDsRed Merge mit mit mit mit

Figure 2 | Expression and localization of Alex3 protein. (a) Anti-Alex3 antibody specifically recognizes Alex3 protein. The antibody recognizes two specific species of about 42 and 39 KDa in HEK293AD cells transfected with Armcx3 cDNA, as shown in WB (left panel). Immunocytochemical detection of Alex3 protein in transfected cells (right panel). (b,c) Coronal sections of adult mouse forebrain showing similar patterns of expression of Armcx3 transcripts (b) and Alex3 protein (c) in the neocortex and hippocampus. (d) Confocal images showing that Alex3 (green) colocalizes with mitochondria in HEK293AD cells; cells were stained with MitoTracker. (e,f) Electron microscopy images of HEK293AD cells transfected with an empty vector (e) or with an Alex3 expression vector (f), illustrating Alex3 localization (gold particles) around mit. (g–j) Immunolocalization of Alex3 protein in cultured hippocampal neurons at 15 DIV. Endogenous Alex3 shows a heterogeneous distribution. Alex3 (green) localizes in nuclei, as shown by colabelling with the nuclear marker Hoechst (blue) (g); in the cytoplasm, Alex3 signals show a punctate pattern that corresponds to the mitochondrial compartment stained with MitoTracker (red) (h); in the cytosol, Alex3 colocalizes with Phalloidin staining (red), particularly in growth cones (arrow) (i). (j) Alex3 staining decorates mitochondria (labelled with MitDSRed) in growth cones. CA1–CA3: Cornu Ammonis1–3; DG: dentate gyrus; I–II–III–IV–V–VI: neocortical layers; mit: mitochondria. Scale bars: 1 mm (b,c); 15 µm (a,c,d); 5 µm (g,h,i,j); and 0.02 µm (e,f). mainly the outer mitochondrial membrane (Fig. 2e,f). These obser- colocalization with the actin cytoskeleton, particularly in neurites vations are consistent with a previous study reporting Alex3 protein and growth cones, where Alex3 also colocalized with mitochondria in outer membrane mitochondrial fractions22. (Fig. 2i,j). Finally, many neurons exhibited Alex3 immunostaining To identify the protein regions required for this targeting, we in nuclei (Fig. 2g). We conclude that endogenous Alex3 is localized generated truncated Alex3–EGFP constructs. Transfection with in three neuronal pools: mitochondria, nucleus and cytosol. deletions at the C-terminus produced a mitochondrial localization pattern that was indistinguishable from that observed with the full- Alex3 regulates mitochondrial network and aggregation. To length Alex3–green fluorescent protein (GFP) cDNA. In contrast, a study whether Alex3 proteins regulate the bioenergetic mitochon- deleted construct at the N-terminus (GFP–Alex3∆1–12) abolished drial state or Ca2 + handling, we performed several experiments. mitochondrial targeting and led to nuclear localization. Moreover, First, we found that overexpression of Alex3 in HEK293T cells did we found that the Arm-containing domains are not required for not alter O2 consumption or DNA copy number (Fig. 3a,b). These mitochondrial targeting (Supplementary Fig. S8b–e). We conclude data, together with findings in neurons showing that Alex3- pro that the N-terminal sequence of Alex3 is necessary and sufficient to tein levels do not modify COX/citrate synthase ratios, DNA copy target the Alex3 protein to the mitochondrial network. number or mitochondrial membrane potential (Fig. 3c–e), indicate We next examined the distribution of Alex3 protein in cul- that this protein is not involved in the regulation of bioenergetic tured hippocampal neurons. These cells exhibited strong cytosolic mitochondrial parameters. To examine whether Alex3 controls Alex3 signals, which colocalized with the mitochondrial network mitochondrial Ca2 + handling26, we transfected cells with a mito- in cell bodies, axons and dendrites (Fig. 2g,h). In addition, Alex3 chondria-targeted aequorin/GFP fusion protein27, and measured signals were distributed throughout the cytosol and showed strong mitochondrial uptake of Ca2 + . Our data show that neither the

a b 3.2 900 2.8 800 cells 6 2.4 10 × 2.0 700 1.6 600 per min 2 1.2 O DNA copy number 0.8 500 nmol 0.4 0 NT pcDNA Alex3 NT pcDNA Alex3 NT pcDNA Alex3 –CCCP +CCCP c d e 150 125 125 125 100 100 100 S 75 75 75 50 50

50 COX/C 25 25 25 DNA copy number (normalized to NI value) (normalized to NI value) 0 0 (normalized to NI value) 0 Knockdown Over Knockdown Over Mit. memb. pot. (FL2/FL1) Knockdown Over expression expression expression

f 7 pcDNA3 Alex3 g pcDNA3 Alex3 6 M 3

µ 1 min

5 M 1 min µ in

M 4 2 ] in 2+ 3 M ]

2 2+ 1 [Ca

1 [Ca 0 0

0 ATP+CCh ATP+CCh 2+ M 2+ 0 M 2+ 0.1 µ 0.1 µ 1 2+ 1 Ca 2+ Ca 2+ Ca Ca Ca Ca

Figure 3 | Neither overexpression nor knockdown of Alex3 impairs the bioenergetic mitochondrial parameters or Ca2 + handling. HEK293T cells (a,b) were transfected with Alex3 or pcDNA3–control, and the oxygen consumption (a) and mitochondrial DNA copy number (b) were measured. There were no differences between control (pcDNA3 or non-transfected (NT) cells) and Alex3-overexpressing cells (a,b). When the uncoupling protein CCCP was added to the medium, oxygen consumption increased but again the differences were not statistically significant. c( ,e) Hippocampal neurons were infected with pWPI (control; black bars) and pWPI–Alex3 (purple bars) in overexpression experiments, and shRNAi–control (black bars) or shRNAi–Alex3 (purple bars) in knockdown experiments. After 7 days, we measured mitochondrial DNA copy number (c), COX and citrate synthase activity (d) and mitochondrial membrane potential (mit. memb. pot.) (e) (JC1 assay, ratio between red (Fl2) and green (FL1) fluorescence). The values were normalized against values from non-infected neurons (NI). No significant differences were found in any of the above parameters. The values show mean ± s.e.m. from three independent experiments. (f,g) Mitochondrial Ca2 + uptake in HEK293T cells transfected with mitochondria-targeted aequorin (see Methods) together with Alex3 (or pcDNA3 in the control). (f) Intact cells were stimulated by ATP + CCh (100 µM each) to trigger maximum IP3-induced Ca2 + release from the ER. Because of extremely close coupling between ER and mitochondria, much of this released Ca2 + is taken up by the mitochondrial Ca2 + uniporter. There were no significant differences between control (pcDNA3) and Alex3 conditions. (g) The mitochondrial uptake from intracellular-like medium was directly measured in cells permeabilized with 60 µM digitonin for 1 min in Ca2 + -free solution (see Methods). Then perfusion was switched first to 0.1 µM Ca2 + , which is insufficient to activate the mitochondrial uniporter, and then to 1µM Ca2 + , which promotes some mitochondrial Ca2 + uptake. Uptake was similar in control and Alex3-transfected cells in all experimental conditions. Data represent mean ± s.e.m. from 4–5 independent experiments. uptake of Ca2 + released from the endoplasmic recticulum (ER) by variable and intermediate degrees of clustered mitochondria (Fig. stimulation with ATP + carbachol nor mitochondrial Ca2 + uptake 4a,c). The severity of the mitochondrial phenotype correlated with by cells whose plasma membrane had been permeabilized by the expression levels of Alex3 protein (Supplementary Fig. S9), in treatment with digitonin was altered by the overexpression of a way similar to that found in Mitofusin 1-overexpressing cells28. Alex3 (Fig. 3f,g). We also imaged Alex3–GFP-transfected cells over 6 h. Concomitant During the above experiments, we noted that cells transfected to Alex3–GFP expression, individual mitochondria aggregated to with Alex3 (as well as with Alex1, 2 and 6 and Armc10) exhib- finally form one or two large clusters Fig.( 4d and Supplementary ited altered mitochondrial networks (Fig. 4a,b and Supplementary Movie 1). Furthermore, electron microscopy analysis confirmed Fig. S6). To further explore this observation, HEK293AD cells were that the overexpression of Alex3 causes the aggregation of mito- transfected with Alex3 and the mitochondrial morphology of these chondria (Fig. 4e,f). cells was monitored 2 days later (Fig. 4a). We found that 88% of To study whether that Alex3 overexpression also leads to mito- these cells presented abnormal, aggregated-like mitochondrial net- chondrial aggregation in neurons, we transfected hippocampal works. Of these cells, 33% showed strongly aggregated phenotypes, cultures (Fig. 4b,c). Up to 85% of Alex3–GFP-transfected neurons in which virtually all the mitochondria were concentrated in large, exhibited an abnormal mitochondrial phenotype; 23% of neurons single aggregates near the nucleus; the remaining 55% showed displayed mitochondria completely aggregated in perinuclear nature communications | 3:814 | DOI: 10.1038/ncomms1829 | www.nature.com/naturecommunications © 2012 Macmillan Publishers Limited. All rights reserved. ARTICLE nature communications | DOI: 10.1038/ncomms1829

0 min 120 min 360 min

a HEK293AD d

b e pcDNA Alex3

Neuron f n nuc nuc s

Normal Mild phenotype Aggregated

c HEK293AD Neurons 100 100

75 75

50 50

25 25 rcentage of cells ercentage of cells P Pe 0 0 pcDNA Alex3 MitDs Alex3 Red GFP

Figure 4 | Alex3 overexpression severely alters mitochondrial morphology and distribution leading to a perinuclear aggregation. (a,b) Overexpression of Alex3 in HEK293AD (a) or Alex3–GFP in 12 DIV hippocampal neurons (b) induces severe alterations in the morphology and distribution of the mitochondrial net when compared with the expression of a control vector (pcDNA in HEK293AD or MitDsRed in neurons). We classified these effects on a 3-point scale from no substantial changes as shown in the first column, a wide spectrum of severe alterations illustrated in the second column and to the most severe phenotype, consisting of the complete aggregation of the mitochondrial net in the perinuclear region, as presented in the third column. (c) Quantification and graphical representation (mean ± s.d.) of the above mentioned subpopulations of Alex3- or ALEX3–GFP-overexpressing cells. We counted 167–345 HEK293 cells and 51–116 neurons per single experiment. Data represent two and three independent experiments, respectively. (d) Series of time-lapsed images of an Alex3–GFP-overexpressing HEK293AD cell showing mitochondrial aggregation. (e,f) Electron microscopy micrographs of HEK293AD cells overexpressing a control vector (e) or Alex3 protein (f). Images show mitochondrial aggregates and their location near the nucleus (d). n: nucleolus; nuc: nucleus. Scale bars: 15 µm (a,b,d); and 2 µm (e,f). clusters, whereas the remaining 62% of neurons displayed inter- mitochondrial aggregation, rather than mitochondrial fusion, by mediate phenotypes with mildly aggregating mitochondria. These mechanisms independent of Miro1. phenotypes were not found in control neurons (Fig. 4c). To confirm that Alex3 causes aggregation, we studied the fine structure of Alex3 regulates mitochondrial trafficking in neurons. Mito- mitochondrial networks in Alex3-transfected hippocampal neurons chondrial transport is essential for neuronal viability18,19,34,35. To (Supplementary Fig. S10). These cells displayed large aggregates of address whether Alex3 is involved in mitochondrial transport in mitochondria in the cell bodies, in which individual mitochondria, axons, we used live-imaging techniques. Cultured hippocampal surrounded by Alex3-immunoreactive end product, were clearly neurons were transfected at 4 DIV with mitochondrial-targeted identifiable. Examination of the neurites also showed the presence DsRed (MitDsRed) or Alex3–GFP. After 2 days, axons were identi- of elongated aggregated mitochondria. Taken together, the above fied and recorded over 15 min (Fig. 6). Control neurons exhibited data suggest that the Armc10/Armcx gene cluster (and particularly high motility rates for mitochondria, which were transported both Armcx3) encodes for proteins involved in mitochondrial aggrega- in the anterograde and retrograde directions (Fig. 6a,b and Sup- tion and dynamics. plementary Movie 6). As described8, only a fraction of individual To directly test whether Alex3 regulates mitochondrial fusion, mitochondria (about 30%, Fig. 6e) moved over the 15-min record- we took advantage of a photoactivable mito-PAGFP version29. ing periods, whereas the remaining ones remained stationary. The HEK293T cells transfected with mito-PAGFP were photoactivated distance covered in single trafficking events by mitochondria was and video recorded over 15 min, and the rates of mitochondrial heterogeneous. While some organelles moved over relatively long fusion were analysed (yellow channel; Fig. 5a,b). In agreement with distances (more than 100–150 µm), others displaced only a few µm previous studies29–31, we observed a constant increase in mitochon- (10–20 µm). The speed at which single mitochondria were trans- drial fusion events over time (Supplementary Movies 2 and 3). The ported was also variable (anterograde: average = 0.21 ± 0.14 µm rate and dynamics of these events in cells transfected with Alex3 per segment; range: 0.07/1.03 µm per segment; retrograde: were identical to those in control cells (Fig. 5c; Supplementary average = 0.29 ± 0.15 µm per segment; range 0.07/0.83 µm per Fig. S11). Similar experiments in hippocampal neurons, either over- segment). Mitochondrial trafficking tended to be more active in the expressing Alex3 or a shRNA sequence, gave similar results (Fig. 5d retrograde than in the anterograde direction (Fig. 6e). and Supplementary Movies 4 and 5). We recorded only Alex3–GFP-tranfected neurons that displayed As Miro1 interacts with Alex3 (see below) and has been shown an apparently normal distribution of mitochondria in the cell bod- to control mitochondrial aggregation32,33, we tested whether this ies and in axons. In these cultures, Alex3–GFP protein was tar- protein mediates the mitochondrial aggregation phenotype induced geted mainly to mitochondria and virtually all the mitochondrial by Alex3. We observed that downregulation of Miro1 protein did meshwork displayed Alex3 signals (Fig. 6a and Supplementary not alter the aggregating phenotype caused by Alex3 (Supplemen- Fig. S13). We also noted that individual mitochondria were larger tary Fig. S12). Taken together, our data indicate that Alex3 regulates in Alex3–GFP-transfected neurons than in control cells (Fig. 6a

a 0 s 450 s 900 s endogenous Alex3 protein. Hippocampal neurons were transfected with pLVTHM vectors expressing shRNAi-Alex3 sequences (Sup- plementary Fig. S16), or a control scrambled sequence. Neurons expressing shRNAi for Alex3 transcripts were viable and morpholog- ically normal, and they extended normal dendrites and axons (Sup- plementary Fig. S17). Alex3–shRNAi-expressing neurons exhibited reduced mitochondrial motility and trafficking and smaller mito- b 0 s 450 s 900 s chondria than control neurons (Fig. 6c,d, Supplementary Movies 8,9, and Supplementary Fig. S14b). This decrease was for anterograde and retrograde transport. However, neither the velocity nor the distance covered by single movements of individual mitochondria was altered by the knockdown of endogenous Alex3 (Fig. 6f). Taken together, these experiments show that Alex3 expression levels regulate the motility of mitochondria in axons in living c 2.5 neurons and modulate both the velocity and distance covered by these organelles. alue) ) 2 Alex3 interacts with Miro and Trak2 proteins. Mitochondrial 37 1.5 trafficking in neurons is mediated by Kinesin (KIF5) motors ,

ed to initial v pcDNA by the Kinesin adaptor Trak2 and by the Rho GTPases Miro1 14,38 otal PDM (+,+

T 1 and Miro2 . We next studied whether Alex3 forms part of the maliz Alex3 KIF5/Miro/Trak2 trafficking complex. First, immunofluorescence

(nor analyses in transfected HEK293AD cells showed strong colocaliza- 0.5 0 150 300 450 600 750 900 tion of Alex3 with Miro1, Miro2 and Trak2, while no apparent colo- Time (s) calization was detected with KIF5C (Supplementary Fig. S18). Next, we performed co-immunoprecipitation assays in HEK293AD cells d 2.5 transfected with Alex3–GFP and either Miro1-myc or Miro2-myc cDNAs. Miro1 and Miro2 were detected by WB in lysates immu- alue) 2 noprecipitated with anti-GFP antibodies. Conversely, pull-downs with anti-myc antibodies revealed Alex3–GFP protein. No co- 1.5 immunoprecipitation was detected when cells were transfected with

ed to initial v pcDNA either Miro1/2-myc or Alex3–GFP DNA alone (Fig. 7a). Similar

otal PDM (+,+) experiments showed that Alex3–GFP was co-immunoprecipitated T 1 Alex3 maliz shRNAi–control with Trak2-myc protein (Fig. 7c). In contrast, we did not detect co-

(nor shRNAi–Alex3 immunoprecipitation of Alex3–GFP with KIF5C, KIF5A or KIF5B 0.5 0 150 300 450 600 750 900 (Fig. 7b). These data indicate that Alex3 interacts directly with the Time (s) Miro1-2/Trak2 proteins, thereby suggesting that this protein is therefore involved in the KIF5/Miro/Trak2 trafficking complex. Figure 5 | Alex3 protein does not alter mitochondrial fusion. To determine whether the Arm-like domains of Alex3 were (a,b) HEK293T cells were transfected with MitDsRed, mito-PAGFP and required for interaction with these proteins, we generated a pcDNA (a), or Alex3 (b). Mito-PAGFP was photoactivated, MitDsRed was N-terminus Alex3(1-106) construct lacking the six Arm domains. photobleached at t = 0 s, and cells were video recorded over 15 min. Panels Co-immunoprecipitation experiments in HEK293T cells showed in (a,b) show the same cell at a range of time points; note the increase that this mutant protein did not co-immunoprecipitate with Trak2 or in yellow signals representing mitochondrial fusion (insets). The rates of Miro2 proteins (Supplementary Fig. S19), thereby indicating that the fusion were analysed (c) using PDM ( + , + ) images (shown in boxed areas Arm-containing C-terminal region is required for this interaction. in (a,b)). Eleven cells, corresponding to two independent experiments, The Miro/Trak2 complex interacts with Kinesin motors in a were quantified for each group. d( ) Rates of mitochondrial fusion over time Ca2 + -dependent manner, in which Ca2 + binding to Miro proteins in hippocampal neurons overexpressing Alex3, or a shRNA, demonstrating triggers the uncoupling of the complex to microtubules, thereby that Alex3 levels do not alter mitochondrial fusion in neurons. At least allowing mitochondrial arrest10,11. We thus tested whether the asso- eight cells, corresponding to two independent experiments, were ciation of Alex3 with the trafficking complex was also 2Ca + -depend- quantified for each group. The values show mean ± s.e.m. Scale bar: 15 µm. ent. The presence of 2 mM Ca2 + markedly reduced the interaction of Alex3 with Miro1, 2 and Trak2 (Fig. 7d,e). Interestingly, cotrans- fection of Alex3–GFP with a Miro1-myc mutant cDNA lacking and Supplementary Fig. S14a). Examination of video recordings the EF-hand domain responsible for Ca2 + binding blocked the indicated that mitochondrial transport was reduced in these neu- regulation of Miro1/Alex3 interaction by Ca2 + , thereby reinforc- rons. To confirm this observation, video recordings were repre- ing the notion that the interaction between these two proteins is sented as kymographs36 and mitochondrial motility was analysed regulated by Ca2 + levels (Fig. 7f). To corroborate these findings, quantitatively (Fig. 6b,e). Overexpression of Alex3–GFP resulted in we purified GST-Miro1 protein and performed pull-downs with a dramatic reduction in the percentage of moving mitochondria, in extracts from brain lysates10. These experiments confirmed the both the anterograde and retrograde directions (Fig. 6e and Sup- direct interaction of Alex3/Miro1 proteins, as well as the regu- plementary Movie 7). This effect was independent of mitochondrial lation of this interaction by Ca2 + and the dependence on the size (Supplementary Fig. S15). Moreover, the velocity and aver- EF-hand domain in Miro1 protein for the regulation of the complex age distance covered by individual mitochondria was impaired in by Ca2 + (Fig. 7g). Taken together, the above data demonstrate that neurons overexpressing Alex3–GFP (Fig. 6e). Alex3 belongs to the KIF5/Miro1-2/Trak2 mitochondrial trafficking To confirm the involvement of Alex3 in mitochondrial dynam- complex and that the Alex3/Miro1 protein interaction requires low ics and trafficking, we performed experiments to knockdown Ca2 + concentrations. nature communications | 3:814 | DOI: 10.1038/ncomms1829 | www.nature.com/naturecommunications © 2012 Macmillan Publishers Limited. All rights reserved. ARTICLE nature communications | DOI: 10.1038/ncomms1829

a c shRNAi–control shRNAi–Alex3

MitDsRed Alex3GFP MitDsRed/GFP 1.5 min 1.5 min

b d n n 15 mi 15 mi Proximal Distal Proximal Distal e f ia ia ia ia

40 25 ) ) 50 50 0.3 m)

0.3 m) –1 60 –1 µ µ 30 s ( ( 30 20 s *** 40 ** 40 m ia ia *** m 0.2 15 0.2 *** µ µ 30 30 * ( 40 20 ( *** 20 10 20 20 *** *** *** *** 0.1 10 0.1 20 10 locity locity Distance per 5 Distance per 10 10 Ve Ve mitochondr 0 0 0.0 mitochondr 0 0 0 0 0 %Motile mitochondr %Motile mitochondr %Motile mitochondr %Motile mitochondr

rade rade rade rade rade rade rade

Retrog Retrog Retrog Retrograde Retrograde Retrograde Anterog Anterog Anterog Anterograde Anterog Anterograde

MitDsRed Alex3GFP shRNAi–control shRNAi–Alex3

Figure 6 | Alex3 protein levels regulate axonal transport of mitochondria in hippocampal neurons. (a,c) Series of four representative confocal images, taken every 30 s, of live axons overexpressing the mitochondrial-tagged protein MitDsRed or Alex3–GFP fusion protein (a), or MitDsRed plus a control shRNAi or a specific Alex3 shRNAi (c). Coloured arrows identify the same mitochondria through the different acquisitions. (c) The upper panel identifies the imaged axon expressing GFP after infection. (b,d) Representative kymographs (two per condition) showing the full-time acquisition periods (that is, 15 min of movie) in overexpressing (c) and in silencing experiments (d). All kymographs are arranged with the distal axonal end to the right. (e,f) Graphical representation of the percentage of motile mitochondria, velocity and distance covered by individual mitochondria measured in kymographs for the overexpression (e) or silencing (f) experiments. Data were analysed using the Student’s t-test and represent the mean ± s.e.m of 22–42 axons (neurons) per experimental group, from at least eight independent experiments. ***P < 0.001; **P < 0.01; *P < 0.05 (For individual P see text). Scale bars: 10 µm.

Discussion including Alex3, have nuclear localization in neurons. This obser- Mitochondrial dynamics is essential for neuronal viability and neu- vation is consistent with a previous study showing interaction rotransmission. Moreover, neuronal mitochondrial dynamics is of Alex3 with the transcription factor Sox10 and enhancement highly coordinated and largely controlled by Kinesins, the GTPases of its transcriptional activity by Alex322. These observations sup- Mitofusins1–2 and Miro1–2, and the adaptor protein Trak23,12,31. port that Armcx genes have biological functions in two compart- Here we describe a family of proteins targeted to mitochondria that ments (namely, mitochondria and nucleus), as has been shown for controls the distribution and dynamics of these organelles and that other proteins, including Sox10, DISC1 and HDAC139–41. Finally, is evolutionarily specific for Eutherian mammals. because of their transcriptional control22 and the presence of six The Armcx gene cluster arose by retrotransposition from a sin- Arm domains, Armcx/Arm10 genes may functionally interact with gle Arm-containing gene (Armc10) exclusive to and present in all the Wnt/β-catenin pathway. The observation that the lack of the vertebrates, and by subsequent short-range tandem duplications of N-terminal region targets Alex3 to the nucleus and the existence a rapidly evolving region of the Eutherian X chromosome. Given of a putative starting codon at the first internal methionine (aa 38) that the expression of retrotransposed genes depends largely on the raise the possibility of alternatively translated isoforms acting selec- regulatory landscape of the insertion site and that gene clusters are tively at the mitochondria or in the nucleus. often maintained to allow the sharing of regulatory sequences, we Our data suggest that Alex3 protein is not involved in mitochon- suggest that the Armcx gene cluster is globally regulated as Armcx drial respiration, DNA copy number determination, mitochondrial gene expression is similar. Furthermore, we hypothesize that the membrane potential regulation, or Ca2 + homoeostasis and han- evolution of this cluster was linked to the increase in the complexity dling in mitochondria. This is interesting because often proteins of the Eutherian brain, thus adding further molecular complexity to regulating mitochondrial dynamics also participate in these bioen- the regulation of mitochondrial trafficking, specifically in the nerv- ergetic processes28. Similarly, we were unable to find evidence of the ous system of higher vertebrates. involvement of Alex3 in mitochondrial fusion. Thus, our data sug- In addition to putative mitochondrial targeting sequences, all gest that the Armcx/Arm10 genes constitute a new family of proteins Armcx/Arm10 genes contain nuclear localization signals and up specifically engaged in mitochondrial dynamics and trafficking and to six Arm repeats at the C-terminal region. Several members, in mitochondrial/nuclear communication.

yc yc yc c

yc yc yc yc rak2my

ak2m a b c Tr 3GFP–T g x3GFP x3GFP–Miro1mx3GFP–Miro2m x3GFP x3GFP–KIF5Cm x3GFP ex GFP Ale GFP–Miro1mAle GFP–Miro2mAle GFP Ale GFP–KIF5CmAle GFP Ale GFP– Al Input c c c 73 110 101 x3 42 my my my Ale Input Input 69 Input 69 69 GFP GFP GFP GST GST–Miro1 2+ c c c

c c 0 50 0 1 5 50 µM Ca c my my my x3 my my my 42 IP IP Ale IP GFP GFP GFP

c GST–Miro1 EF

c c ∆ P 2+ my my my 0 1 5 50 µM Ca x3 IP GFP IP GFP IP GF 42 GFP GFP GFP Ale

yc c yc yc EFm yc yc ∆ my 2m yc yc yc yc ak EFm Tr 1∆ ak2m Tr d x3GFP x3GFP–Miro1mx3GFP–Miro2me x3GFP x3GFP–Miro1x3GFP–Miro1 f x3GFP x3GFP– GFP Ale GFP–Miro1mAle GFP–Miro2mAle GFP Ale GFP–MiroAle GFP–Miro1mAle GFP Ale GFP– Ale c c c 73 73 101 my my my Input Input Input 69 69 69 GFP GFP GFP 2+ – + – + – + – + – + – + Ca2+ – + – + – + – + – + – + Ca2+ – + – + – + – + Ca c c c c c c my my my my my my IP IP IP GFP GFP GFP c c c my my my IP GFP IP GFP IP GFP GFP GFP GFP

Figure 7 | Alex3 is a component of the protein complex responsible for mitochondrial transport. (a–f) A GFP construct was used as control. (a–c) Co-immunoprecipitation of Alex3–GFP with Miro1myc, Miro2myc, KIF5Cmyc or Trak2myc results in the specific interaction of Alex3–GFP with Miro1myc, Miro2myc (a) and Trak2myc (c) but no interaction is detected between Alex3–GFP and KIF5Cmyc (b). (d) The interaction between Alex3– GFP and Miro1myc and Miro2myc is reduced in the presence of 2 mM of calcium. (e) When the EF-hand mutant form of Miro1myc is co-expressed with Alex3–GFP, the interaction between the two components is no longer sensitive to Ca2 + levels. (f) The association between Alex3–GFP and Trak2myc is also dependent on the presence of Ca2 + , thereby indicating that the whole complex could be dismantled when Miro1/2 recognizes an increment in Ca2 + levels. (g) Pull-down assays using purified GST–Miro1 and GST–Miro1∆EF proteins as bait and showing interaction with Alex3 (in brain lysates) and its dependence on Ca2 + concentration (top panel). The lack of the EF-hand domain in Miro-1 protein abolishes Ca2 + dependence (bottom panel).

Alex3 causes mitochondrial aggregation and/or tethering, and mutation of the EF-hand motif in Miro1 protein abolished this this effect is independent of the levels of Miro1, a protein involved Ca2 + dependence, thereby indicating that Ca2 + -driven conforma- in mitochondrial aggregation32,33. These findings suggest that tional changes in Miro proteins10,11,46 are the essential mechanisms Alex3 regulates aggregation by an as yet unknown mechanism, that regulate the interaction between Alex3 and the Miro/Trak2 which is likely to involve interaction with proteins that control complex. Thus, while low 2 Ca + concentrations may favour the tethering and aggregation, including Mitofusins28. It has been formation of KIF5/Miro/Trak2/Alex3 complexes, increases in intra- reported that Arm motifs of p115 protein bind the Rab1 GTPase cellular Ca2 + rapidly uncouple such complexes (including Alex3), controlling ER to Golgi trafficking and regulate vesicle tethering42. thereby arresting mitochondrial trafficking. Mitochondrial aggregating phenotypes have been observed after Our experiments indicate that the C-terminal region containing dysfunction of Miro and Trak2 proteins, suggesting that alteration six Arm domains is required to interact with Miro protein. However, of transport results in aggregation43,44. In neurons, mitochondrial we do not yet know whether this interaction is mediated directly by aggregation and tethering is believed to serve to capture these these domains, as is the case for other proteins47, or by other regions organelles at specific locations requiring high-energy and Ca2 + included in the C-terminus. buffering conditions43,45. The notion that Alex3 interacts with the Miro/Trak2 complex At least one of the Armcx genes (Alex3) interacts with the KIF5/ when mitochondria are motile at low Ca2 + concentrations10,11 is Miro/Trak2 complex, which controls mitochondrial dynamics in further supported by our findings that knockdown of Alex3 results neurons in a Ca2 + -dependent manner. Our immunoprecipitation in a decrease in the percentages of motile mitochondria, similarly experiments show that Alex3 interacts directly with Miro1-2 and to what was observed in Miro/Trak2 loss-of-function10,31,48. The Trak2. However, we were unable to find a direct interaction with observation that knockdown of Alex3 does not affect the speed the Kinesin motor KIF5. Importantly, the interaction of Alex3 with of the few motile mitochondria present suggests a mechanism in Miro/Trak2 proteins requires low Ca2 + concentrations, as the pres- which Alex3 favours the formation of KIF5/Miro/Trak2 complexes ence of Ca2 + dramatically reduced the interaction. Furthermore, (thereby enhancing mitochondrial trafficking). However, Alex3 is nature communications | 3:814 | DOI: 10.1038/ncomms1829 | www.nature.com/naturecommunications © 2012 Macmillan Publishers Limited. All rights reserved. ARTICLE nature communications | DOI: 10.1038/ncomms1829 unlikely to be involved in regulating the motor activity of Kinesin Cell culture and bioenergetic parameters. HEK293AD cells were used for all itself. We consider our Alex3 overexpression findings, in which both the experiments. Cells were cultured in DMEM medium supplemented with 10% − 1 − 1 the percentage and velocity of motile mitochondria were reduced, fetal bovine serum, 2 mM glutamine, 120 µg ml Penicillin and 200 µg ml Streptomycin. Upon confluence, cells were trypsinized (0.25% w/v) and plated to provide further evidence of the physiological involvement of at the desired density. After 2 days, cells were transfected using Fugene6 (Roche this protein in mitochondrial trafficking, possibly by deregulating Diagnostics), following the manufacturer’s instructions, and using a 1:1 DNA ratio and/or recruiting components of the complex (acting as a dominant when two constructs were transfected. For immunoprecipitation assays, an empty negative). Together with our biochemical data, the present func- vector was used to balance transfected DNA when required. tional study proposes a model in which Alex3 protein is a positive E16 mouse brains were dissected in PBS containing 0.6% glucose and the hippocampi were dissected out. After trypsin (Invitrogen) and DNAse treatment regulator of mitochondrial trafficking, by interacting directly with (Roche Diagnostics), tissue pieces were dissociated, and cells were seeded onto Miro/Trak2. Furthermore, as shown for the KIF5/Miro/Trak2 com- 0.5 mg ml − 1 poly-l-lysine (Sigma-Aldrich)-coated coverslips (for immunocyto- plex10,11, increased neuronal activity leading to increases in Ca2 + is chemistry) or 35-mm Fluorodish plates (World Precision Instruments, Inc) for live imaging in neurobasal medium (Gibco) containing 2 mM glutamax, 120 µg ml − 1 likely to cause Alex3/Miro/Trak2 complex disassembly and mito- − 1 chondrial arrest at active neurotransmission sites, thereby fulfilling Penicillin, 200 µg ml Streptomycin and B27 supplement (Invitrogen), and were maintained at 37 °C in the presence of 5% CO2. Cells were cultured for between 7 the bioenergetic requirements of neuronal transmission. and 15 days. Neuron transfection was carried out at 10–12 DIV for immunocyto- In conclusion, here we describe a new family of genes organized chemistry experiments and at 4 DIV for live imaging, using Lipofectamine 2000 in a genomic cluster encoding proteins targeted to mitochondria (Invitrogen), following the manufacturer’s instructions, and using a 1:3 DNA ratio that regulates mitochondrial dynamics and trafficking in neurons, when transfection of two constructs was required. Cells were processed 24–48 h after transfection. Neurite extension and branching was analysed as described50 and interacts with the KIF5/Miro/Trak2 complex. Importantly, in three experimental groups (control–GFP, Alex3–GFP/m-Cherry and shR- the Armcx cluster emerged suddenly during the early evolution of NAi–Alex3) Immunocytochemical and electron microscopy techniques as well Eutheria. We thus propose a further complexity in the molecular as methods used to determine bioenergetic parameters and Ca2 + handling are mechanisms controlling mitochondrial dynamics, specifically in detailed in Supplementary Methods. the neurons of more evolved brains. Our data point to relevant Imaging and quantification of axonal transport of mitochondria. HEK293AD functions of the Armcx gene cluster in the function of the nervous cells and hippocampal neurons were seeded onto Poly-l-lysine-coated Fluorodish system and a novel link between mitochondrial dynamics and brain plates (World Precision Instruments, Inc), transfected with Alex3–GFP, MitDsRed complexity. or mito-PAGFP and filmed 24 to 48 h after using a Leica TCS SP2 confocal micro- scope (Leica Microsystems) equipped with a ×63 immersion oil objective. Methods Methods used to determine mitochondrial fusion are detailed in Supplementary Methods. DNA analyses. Armcx DNA sequences, protein translation and chromosomal For mitochondrial aggregation in transfected HEK293AD cells, time lapse series of linkage analyses were performed by browsing public genome data (www.ensembl. image stacks composed of 10 images (512×512 px) were taken every 5 min over 6 h org, www.ncbi.nlm.nih.gov) and manual curation through analysis and BlastN using Leica Confocal Software (Leica Microsystems). Movies were generated at 10 of EST databases in NCBI as well as by reciprocal blasting between mammalian frames per second. For mitochondrial fusion analyses in HEK293T cells, time lapse orthologs. Domain structure was determined by NCBI, Prosite, InterProScan, series of image stacks composed of seven images (512×512 px) were taken every 6 s IPSort, HHPred and Wolfsort software. Phylogenetic trees were made by Bayesian over 15 min. Movies were generated at 7 frames per second. Axonal mitochondria methods using MrBayes 3.1.249. were registered with an additional digital zoom of ×1.7. Time lapse series of image stacks composed of 5 images (512×512 px) were taken every 6 s during 15 min. All Plasmid vectors. Alex3 full sequence was obtained by screening a P0 mouse 151 stacks obtained were processed mainly with Leica Confocal Software. Further brain cDNA library (Stratagene). For the generation of Alex3, Alex3–myc and image processing, analysis and video compilation (10 frames per second) and GFP–Alex3∆(1–12) expression vectors, pBluescript–Alex3 was subcloned into the edition was done with ImageJ software (version 1.43K, NIH, USA). Kymographs desired expression vectors: pcDNA.3 (Invitrogen, Carlsbad, CA, USA), pSecTag-A were generated with MetaMorph Software (Molecular Devices–MDS Analyti- (Invitrogen), and pEGFP-N3 and pEGFP-C1 (Clontech). For the generation of cal Technologies). In overexpression experiments, 42 axons were registered and Alex3–GFP, Alex3(1–200)–GFP, Alex3(1–106)–GFP and Alex3(1–45)–GFP , Alex3 analysed in each condition while in silencing experiments 22 axons per group were was amplified with high-fidelity Pfu, and a BamHI restriction site was introduced recorded. In all cases, a mitochondrion was considered motile when it moved more µ by using appropriate primers (Forward: 5′-CTATAGGGCGAATTGGGTACCG-3′; than 0.5 m during 1 min of recording. Distances and speeds of retrograde and anterograde transport were measured separately from the corresponding and reverse: for Alex3–GFP 5′-GGATCCTTCCTGACTCTTTGGGAACATCC-3′; 51 for Alex3(1–200)–GFP 5′-GGATCCAAGCCTGCGCTGATTTTCGG-3′ and for kymographs, as previously described , and no tracking–plugging was used. Im- Alex3(1–45)–GFP 5′-GGATCCATCACCAGAGCCACCCTCA-3′). All the con- ageJ software was used to quantify mitochondrial length. The first frame recorded structs were sequenced with BigDye-Terminator v3.1 (Applied Biosystems). shR- of each video was digitally processed and thresholded (using Yi algorithm), and NAi sequences were designed against the Alex3 messenger RNA sequence using the longest distance between any two points along the selection boundary (Feret’s siRNA primer designer software v1.51 (Promega) and subcloned into pLVTHM parameter) was measured for all mitochondria. To ensure equal relevance between vector (Tronolab) through MluI/ClaI (New England Biolabs) digestion. The seq axons, independently of the number of mitochondria displayed, an average length uence of the selected shRNAi–Alex3 was 5′-CGCGTCCCCGGCTTTAATTGT was calculated for each neuron and then the mean was calculated between axons. CCTAAATTTCAAGAGAATTTAGGACAATTAAAGCCTTTTTGGAAAT-3′; and the scrambled shRNAi–control 5′-CGCGTCCCCGCAGCTTTAATTGCTC Protein studies. HEK293AD cells were obtained and lysed in Laemmli Buffer at TGGGTAATAATATTCATATTACCCAGAGCAATTAAAGCTGCTTTTTGG 98 °C for 5 min. Brain extracts were obtained by homogenization in lysis buffer AAAT-3′. The expression vectors Miro1–myc, Miro2–myc, Miro1∆EF–myc, (50 mM HEPES; 150 mM NaCl; 1.5 mM MgCl2; 1 mM EGTA; 10% glycerol; KIF1C–myc and Trak2–myc were as previously described10,38. Miro1 shRNAi 1% Triton X-100; protease inhibitor cocktail (Roche Diagnostics Gmbh), 1 mM (V2MM_72258, TGCTGTTGACAGTGAGCGCCCTGATTGCACTACTGAATTA NaF, 0.5 M sodium pyrophosphate and 200 mM orthovanadate). About 20 µg of TAGTGAAGCCACAGATGTA TAATTCAGTAGTGCAATCAGGATGCCTACT protein for each sample was loaded and run in polyacrylamide gels at 100 V. Trans- GCCTCGGA; and a control scrambled sequence) was provided by Josef Kittler fer to nitrocellulose membranes was performed in 120 mM glycine, 125 mM Tris, (UCL, London). The mitochondrial-targeted aequorin and mito-PAGFP were as 0.1% SDS and 20% methanol at 35V over night. Membranes were then blocked previously described27,29. MitDsRed was a gift from Antonio Zorzano (IRB Barce- in 5% powder milk in TBS and incubated with primary antibodies anti-Alex3 lona). GST–Miro1 and GST–Miro1∆EF were kindly provided by Josef Kittler10. (1:2000), anti-GFP (1:1000, Invitrogen), anti-myc (1:2000, Santa Cruz Biotechnolo- gies) or anti-KHC (1:1000, Millipore). Anti-actin was used as a loading control Animals. All procedures were performed in accordance with the guidelines (1:1000, Chemicon, Temecula, CA, USA). Secondary antibodies coupled to horse- approved by the Spanish Ministry of Science and Technology and following the radish peroxidase were used diluted 1:2500 in TBS containing 5% powder milk. European Community Council Directive 86/609 EEC. OF1 embryos and postnatal Labelling was visualized with ECL plus (Amersham Pharmacia Biotech). mice (Iffra Credo, Lyon, France) were used. The mating day was considered embry- HEK293AD cells were lysed using a buffer containing 50 mM Tris–HCl pH7.5; onic day 0 (E0) and the day of birth postnatal day 0 (P0). The following develop- 150 mM NaCl; 1.5mM MgCl2; 5 mM EDTA; 1% Triton-100; 10% glycerol; protease mental stages were studied: E9, E10, E11, E12, E16, P6 P21 and adult. inhibitor cocktail (Roche Diagnostics Gmbh); 1 mM NaF; 0.5 M sodium pyro- Animals were anesthetized with 4% halothane before killing. Histological and in phosphate and 200 mM orthovanadate. About 500 µg of total protein per sample situ hybridization techniques are detailed in Supplementary Methods. was used for the immunoprecipitation assays. Homogenates were incubated with anti-GFP (1:500, Invitrogen) and anti-myc (1:500, Santa Cruz Biotechnology), at Alex3 antibodies. Polyclonal antibodies against Alex3 were generated by injecting 4 °C o.n. overnight. All the assays were performed using protein G-Sepharose beads rabbits with a synthesized peptide (369-LTERMFPKSQE-379). (Sigma-Aldrich) for 2 h at 4 °C. The assays performed in the presence of Ca2 + ,

10 nature communications | 3:814 | DOI: 10.1038/ncomms1829 | www.nature.com/naturecommunications © 2012 Macmillan Publishers Limited. All rights reserved. nature communications | DOI: 10.1038/ncomms1829 ARTICLE

EDTA and EGTA were omitted from the lysis buffer and Ca2 + was added to a final 24. Simonin, F., Karcher, P., Boeuf, J. J., Matifas, A. & Kieffer, B. L. Identification concentration of 2 mM. After incubation with protein G-Sepharose, samples were of a novel family of G protein-coupled receptor associated sorting proteins. J. washed five times in washing buffer (10 mM Tris–HCl pH8, 500 mM NaCl, 1 mM Neurochem. 89, 766–775 (2004). EDTA, 1 mM EGTA, 1% Triton X-100 and 0.5% NP-40) in which EDTA and EGTA 25. Puelles, L. Brain segmentation and forebrain development in amniotes. Brain were again substituted by 2 mM Ca2 + when required. Laemmli Buffer was added to Res. Bull. 55, 695–710 (2001). the beads and they were then boiled at 98 °C for 5 min. Proteins were then analysed 26. Singaravelu, K. et al. Mitofusin 2 regulates STIM1 migration from the Ca2+ by SDS–PAGE and WB. store to the plasma membrane in cells with depolarized mitochondria. J. Biol. Chem. 286, 12189–12201 (2011). Miro1-binding assay. GST fusion proteins (GST–Miro1 and GST–Miro1∆ EF) 27. Manjarres, I. M. et al. Red and green aequorins for simultaneous monitoring of were produced in E. coli and purified as described52. Brain lysates (adult brain) Ca2+ signals from two different organelles.Pflugers Arch. 455, 961–970 (2008). were solubilized in a buffer containing 50 mM HEPES, 125 mM NaCl, 1% Triton, 28. Santel, A. & Fuller, M. T. Control of mitochondrial morphology by a human 2 mM EDTA, 3 mM EGTA, 1 mM PMSF and antipain, pepstatin and leupeptin at mitofusin. J. Cell Sci. 114, 867–874 (2001). 10 mg ml − 1. GST fusion proteins coupled to Sepharose beads (30 µg) was exposed 29. Karbowski, M. et al. Quantitation of mitochondrial dynamics by photolabeling to 20 mg of brain lysate with varying concentrations of [Ca2 + ]. Bound material was of individual organelles shows that mitochondrial fusion is blocked during the washed five times in the above buffer before elution with SDS sample buffer. WB Bax activation phase of apoptosis. J. Cell Biol. 164, 493–499 (2004). was carried as described above. 30. Gomes, L. C., Di Benedetto, G. & Scorrano, L. During autophagy mitochondria elongate, are spared from degradation and sustain cell viability. Nat. Cell Biol. Statistical analysis. Data were analysed with the Igor Pro (WaveMetrics) and 13, 589–598 (2011). Origin 5.0. (Microcal Software) using the Student’st test or the Mann–Whitney– 31. Saotome, M. et al. Bidirectional Ca2+-dependent control of mitochondrial Wilcoxon test for non-parametric data. Statistical significance was fixed atP < 0.05; dynamics by the Miro GTPase. Proc. Natl Acad. Sci. USA 105, 20728–20733 (2008). * indicates P < 0.05, **P < 0.01 and ***P < 0.001. 32. Fransson, S., Ruusala, A. & Aspenstrom, P. The atypical Rho GTPases Miro-1 and Miro-2 have essential roles in mitochondrial trafficking.Biochem. Biophys. Res. Commun. 344, 500–510 (2006). References 33. Fransson, A., Ruusala, A. & Aspenstrom, P. Atypical Rho GTPases have roles 1. Hollenbeck, P. J. Mitochondria and neurotransmission: evacuating the synapse. in mitochondrial homeostasis and apoptosis. J. Biol. Chem. 278, 6495–6502 Neuron 47, 331–333 (2005). (2003). 2. Ly, C. V. & Verstreken, P. Mitochondria at the synapse. Neuroscientist 12, 34. Delettre, C. a mitochondrial dynamin-related protein, is mutated in dominant 291–299 (2006). optic atrophy. Nat. Genet. 26, 207–210 (2000). 3. Verstreken, P. et al. Synaptic mitochondria are critical for mobilization of 35. Alexander, C. et al. OPA1, encoding a dynamin-related GTPase, is mutated in reserve pool vesicles at Drosophila neuromuscular junctions. Neuron 47, autosomal dominant optic atrophy linked to chromosome 3q28. Nat. Genet. 26, 365–378 (2005). 211–215 (2000). 4. Ferreirinha, F. et al. Axonal degeneration in paraplegin-deficient mice is 36. Miller, K. E. & Sheetz, M. P. Axonal mitochondrial transport and potential are associated with abnormal mitochondria and impairment of axonal transport. correlated. J. Cell Sci. 117, 2791–2804 (2004). J. Clin. Invest. 113, 231–242 (2004). 37. Hirokawa, N., Noda, Y., Tanaka, Y. & Niwa, S. Kinesin superfamily motor 5. Lee, C. W. & Peng, H. B. The function of mitochondria in presynaptic proteins and intracellular transport. Nat. Rev. Mol. Cell. Biol. 10, 682–696 (2009). development at the neuromuscular junction. Mol. Biol. Cell 19, 150–158 (2008). 38. MacAskill, A. F., Brickley, K., Stephenson, F. A. & Kittler, J. T. GTPase 6. Chan, D. C. Mitochondrial fusion and fission in mammals.Ann. Rev. Cell Dev. dependent recruitment of Grif-1 by Miro1 regulates mitochondrial trafficking Biol. 22, 79–99 (2006). in hippocampal neurons. Mol. Cell. Neurosci. 40, 301–312 (2009). 7. Cai, Q. & Sheng, Z. H. Moving or stopping mitochondria: Miro as a traffic cop 39. Atkin, T. A., MacAskill, A. F., Brandon, N. J. & Kittler, J. T. Disrupted in by sensing calcium. Neuron 61, 493–496 (2009). Schizophrenia-1 regulates intracellular trafficking of mitochondria in neurons. 8. Hollenbeck, P. J. & Saxton, W. M. The axonal transport of mitochondria.J. Cell Mol. Psychiatry 16, 122–124, 121 (2011). Sci. 118, 5411–5419 (2005). 40. Kim, J. Y. et al. HDAC1 nuclear export induced by pathological conditions is 9. Hollenbeck, P. J. The pattern and mechanism of mitochondrial transport in essential for the onset of axonal damage. Nat. Neurosci. 13, 180–189 (2010). axons. Front Biosci. 1, d91–102 (1996). 41. Sawamura, N. et al. Nuclear DISC1 regulates CRE-mediated gene transcription 10. Macaskill, A. F. et al. Miro1 is a calcium sensor for glutamate receptor- and sleep homeostasis in the fruit fly.Mol. Psychiatry 13, 1138–1148, 1069 (2008). dependent localization of mitochondria at synapses. Neuron 61, 541–555 42. An, Y. et al. Structural and functional analysis of the globular head domain of (2009). p115 provides insight into membrane tethering. J. Mol. Biol. 391, 26–41 (2009). 11. Wang, X. & Schwarz, T. L. The mechanism of Ca2+ -dependent regulation of 43. MacAskill, A. F. & Kittler, J. T. Control of mitochondrial transport and kinesin-mediated mitochondrial motility. Cell 136, 163–174 (2009). localization in neurons. Trends Cell Biol. 20, 102–112 (2010). 12. Misko, A., Jiang, S., Wegorzewska, I., Milbrandt, J. & Baloh, R. H. Mitofusin 2 44. Liu, X. & Hajnoczky, G. Ca2+-dependent regulation of mitochondrial is necessary for transport of axonal mitochondria and interacts with the Miro/ dynamics by the Miro-Milton complex. Int. J. Biochem. Cell Biol. 41, 1972–1976 Milton complex. J. Neurosci. 30, 4232–4240 (2010). (2009). 13. Hoppins, S., Lackner, L. & Nunnari, J. The machines that divide and fuse 45. Chang, D. T. & Reynolds, I. J. Mitochondrial trafficking and morphology in mitochondria. Ann. Rev. Biochem. 76, 751–780 (2007). healthy and injured neurons. Prog. Neurobiol. 80, 241–268 (2006). 14. Guo, X. et al. The GTPase dMiro is required for axonal transport of 46. Nelson, M. R. & Chazin, W. J. Structures of EF-hand Ca(2+)-binding proteins: mitochondria to Drosophila synapses. Neuron 47, 379–393 (2005). diversity in the organization, packing and response to Ca2+ binding. Biometals 15. Pilling, A. D., Horiuchi, D., Lively, C. M. & Saxton, W. M. Kinesin-1 and 11, 297–318 (1998). Dynein are the primary motors for fast transport of mitochondria in 47. Tewari, R., Bailes, E., Bunting, K. A. & Coates, J. C. Armadillo-repeat protein Drosophila motor axons. Mol. Biol. Cell 17, 2057–2068 (2006). functions: questions for little creatures. Trends Cell Biol. 20, 470–481 (2010). 16. Stowers, R. S., Megeath, L. J., Gorska-Andrzejak, J., Meinertzhagen, I. A. & 48. Brickley, K. & Stephenson, F. A. Trafficking kinesin protein (TRAK)-mediated Schwarz, T. L. Axonal transport of mitochondria to synapses depends on transport of mitochondria in axons of hippocampal neurons. J. Biol. Chem. Milton, a novel Drosophila protein. Neuron 36, 1063–1077 (2002). 286, 18079–18092 (2011). 17. Chen, H. & Chan, D. C. Critical dependence of neurons on mitochondrial 49. Huelsenbeck, J. P. & Ronquist, F. MRBAYES: Bayesian inference of phylogenetic dynamics. Curr. Opin. Cell Biol. 18, 453–459 (2006). trees. Bioinformatics (Oxford, England) 17, 754–755 (2001). 18. Zuchner, S. et al. Mutations in the mitochondrial GTPase mitofusin 2 cause 50. Vitureira, N. et al. Podocalyxin is a novel polysialylated neural adhesion protein Charcot-Marie-Tooth neuropathy type 2A. Nat. Genet. 36, 449–451 (2004). with multiple roles in neural development and synapse formation. PloS One 5, 19. Zhao, C. et al. Charcot-marie-tooth disease type 2A caused by mutation in a e12003 (2010). microtubule motor KIF1Bbeta. Cell. 105, 587–597 (2001). 51. De Vos, K. J. & Sheetz, M. P. Visualization and quantification of mitochondrial 20. Kurochkin, I. V., Yonemitsu, N., Funahashi, S. I. & Nomura, H. ALEX1, a dynamics in living animal cells. Methods Cell Biol. 80, 627–682 (2007). novel human armadillo repeat protein that is expressed differentially in normal 52. Kittler, J. T., Hanley, J. G. & Isaac, J. T. R. The Dynamic Synapse: Molecular tissues and carcinomas. Biochem. Biophys. Res. Commun. 280, 340–347 (2001). Methods in Ionotropic Receptor Biology, Chapter 11 (Boca Raton (FL), CRC 21. Garcia-Frigola, C. et al. A collection of cDNAs enriched in upper cortical layers Press 2006. of the embryonic mouse brain. Brain Res. 122, 133–150 (2004). 22. Mou, Z., Tapper, A. R. & Gardner, P. D. The armadillo repeat-containing Acknowledgements protein, ARMCX3, physically and functionally interacts with the developmental This project was supported by grant BFU2008-3980 (Ministerio de Ciencia e Innovacion regulatory factor Sox10. J. Biol. Chem. 284, 13629–13640 (2009). (MICINN), Spain) and grants from the ‘Marató de TV3’ and ‘Caixa Catalunya-Obra 23. Winter, E. E. & Ponting, C. P. Mammalian BEX, WEX and GASP genes: coding Social’ Foundations to E.S., by grant BFU2011-23921 (MICINN) and the ICREA and non-coding chimaerism sustained by gene conversion events. BMC Evol. Academia Prize (Generalitat de Catalunya) to J.G.-F., by grants PI07/0715 (Fondo de Biol. 5, 54 (2005). Investigación Sanitaria) to F.B., by grant SAF2008-03514 to R.T. and by grants from the nature communications | 3:814 | DOI: 10.1038/ncomms1829 | www.nature.com/naturecommunications 11 © 2012 Macmillan Publishers Limited. All rights reserved. ARTICLE nature communications | DOI: 10.1038/ncomms1829

EU-ERA-Net program and MICINN (SAF2008-03175-E, BFU2010-17379 and Red de Additional information Terapia Celular, RD06/0010/0000) to J.G.-S. I.S. holds an IEF Postdoctoral fellowship Supplementary Information accompanies this paper at http://www.nature.com/ funded by the Marie Curie FP7 ‘People’ Programme. We thank Dr J Kittler (UCL, naturecommunications London) and Dr Antonio Zorzano (IRB Barcelona, Spain) for generously providing material, Carles Bosch for electron microscopy and Tanya Yates for editorial help. Competing financial interests: The authors declare no competing financial interests. Reprints and permission information is available online at http://npg.nature.com/ Author contributions reprintsandpermissions/ G.L.-D., R.S., S.M. and E.S. designed and performed cellular, histological, imaging and How to cite this article: López-Doménech, G. et al. The EutherianArmcx genes regulate molecular experiments, analysed the data and interpreted the experimental results. S.D. mitochondrial trafficking in neurons and interact with Miro and Trak2.Nat. Commun. and I.S. performed genetic analysis and in situ hybridizations; A.A., N.V., E.G.-A., A.L.A., 3:814 doi: 10.1038/ncomms1829 (2012). R.T., F.B., J.G.-S, M.T.A and M.R.-P. performed the experiments and analysed the data; License: This work is licensed under a Creative Commons Attribution-NonCommercial- J.G.-F supervised the genetic studies; J.G.-S., J.G.-F. and E.S. wrote the manuscript; and NoDerivative Works 3.0 Unported License. To view a copy of this license, visit http:// E.S. supervised the project. creativecommons.org/licenses/by-nc-nd/3.0/